bool SampleFile::takeSample()
{
//we're only operating on one file, so the idx is zero.
_inputFile = new FileRecordMgr(0, _context);
if (!_inputFile->open()) {
return false; // FRM will give the error and die anyway,
//no error message needed here.
}
_samples.resize(_numSamples, NULL);
//Context object takes care of the seed, either user given or randomly
//generated, and seeds the call to srand with it, so we don't have to
//here.
if (!_context->hasConstantSeed()) {
_context->getUnspecifiedSeed();
}
_context->determineOutputType();
_outputMgr = new RecordOutputMgr();
_outputMgr->init(_context);
while (!_inputFile->eof()) {
Record *record = _inputFile->allocateAndGetNextRecord();
if (record == NULL) {
continue;
}
if (!keepRecord(record)) {
_inputFile->deleteRecord(record);
}
_currRecordNum++;
}
if (_currRecordNum < _numSamples) {
//die with error;
cerr << "\n***** ERROR: Input file has fewer records than the requested number of output records. *****" << endl << endl;
exit(1);
}
//If the output type is BAM, must sort the output records.
if (_context->getOutputFileType() == FileRecordTypeChecker::BAM_FILE_TYPE) {
sort(_samples.begin(), _samples.end(), SampleRecordLtFn);
}
// Now output all the kept records, then do cleanup.
for (size_t i=0; i < _numSamples; i++) {
_outputMgr->printRecord(_samples[i]);
}
delete _outputMgr;
_inputFile->close();
//clean up.
delete _inputFile;
return true;
}
bool SampleFile::findNext(RecordKeyVector &hits) {
while (!_inputFile->eof()) {
Record *record = _inputFile->getNextRecord();
if (record == NULL) {
continue;
} else {
_currRecordNum++;
if (!keepRecord(record)) {
_inputFile->deleteRecord(record);
}
return true;
}
}
return false;
}
